Merge remote-tracking branch 'upstream/master' into SPARK-16004

Author: bomeng

core/src/main/scala/org/apache/spark/SparkContext.scala

@@ -391,7 +391,7 @@ class SparkContext(config: SparkConf) extends Logging with ExecutorAllocationCli
 
     _conf.set("spark.executor.id", SparkContext.DRIVER_IDENTIFIER)
 
-    _jars = _conf.getOption("spark.jars").map(_.split(",")).map(_.filter(_.nonEmpty)).toSeq.flatten
+    _jars = Utils.getUserJars(_conf)
     _files = _conf.getOption("spark.files").map(_.split(",")).map(_.filter(_.nonEmpty))
       .toSeq.flatten
 

core/src/main/scala/org/apache/spark/util/Utils.scala

@@ -2352,6 +2352,31 @@ private[spark] object Utils extends Logging {
     log.info(s"Started daemon with process name: ${Utils.getProcessName()}")
     SignalUtils.registerLogger(log)
   }
+
+  /**
+   * Unions two comma-separated lists of files and filters out empty strings.
+   */
+  def unionFileLists(leftList: Option[String], rightList: Option[String]): Set[String] = {
+    var allFiles = Set[String]()
+    leftList.foreach { value => allFiles ++= value.split(",") }
+    rightList.foreach { value => allFiles ++= value.split(",") }
+    allFiles.filter { _.nonEmpty }
+  }
+
+  /**
+   * In YARN mode this method returns a union of the jar files pointed by "spark.jars" and the
+   * "spark.yarn.dist.jars" properties, while in other modes it returns the jar files pointed by
+   * only the "spark.jars" property.
+   */
+  def getUserJars(conf: SparkConf): Seq[String] = {
+    val sparkJars = conf.getOption("spark.jars")
+    if (conf.get("spark.master") == "yarn") {
+      val yarnJars = conf.getOption("spark.yarn.dist.jars")
+      unionFileLists(sparkJars, yarnJars).toSeq
+    } else {
+      sparkJars.map(_.split(",")).map(_.filter(_.nonEmpty)).toSeq.flatten
+    }
+  }
 }
 
 /**

core/src/test/scala/org/apache/spark/deploy/SparkSubmitSuite.scala

@@ -570,6 +570,18 @@ class SparkSubmitSuite
       appArgs.executorMemory should be ("2.3g")
     }
   }
+
+  test("comma separated list of files are unioned correctly") {
+    val left = Option("/tmp/a.jar,/tmp/b.jar")
+    val right = Option("/tmp/c.jar,/tmp/a.jar")
+    val emptyString = Option("")
+    Utils.unionFileLists(left, right) should be (Set("/tmp/a.jar", "/tmp/b.jar", "/tmp/c.jar"))
+    Utils.unionFileLists(emptyString, emptyString) should be (Set.empty)
+    Utils.unionFileLists(Option("/tmp/a.jar"), emptyString) should be (Set("/tmp/a.jar"))
+    Utils.unionFileLists(emptyString, Option("/tmp/a.jar")) should be (Set("/tmp/a.jar"))
+    Utils.unionFileLists(None, Option("/tmp/a.jar")) should be (Set("/tmp/a.jar"))
+    Utils.unionFileLists(Option("/tmp/a.jar"), None) should be (Set("/tmp/a.jar"))
+  }
   // scalastyle:on println
 
   // NOTE: This is an expensive operation in terms of time (10 seconds+). Use sparingly.

data/mllib/sample_isotonic_regression_data.txt

@@ -0,0 +1,100 @@
+0.24579296 1:0.01
+0.28505864 1:0.02
+0.31208567 1:0.03
+0.35900051 1:0.04
+0.35747068 1:0.05
+0.16675166 1:0.06
+0.17491076 1:0.07
+0.04181540 1:0.08
+0.04793473 1:0.09
+0.03926568 1:0.10
+0.12952575 1:0.11
+0.00000000 1:0.12
+0.01376849 1:0.13
+0.13105558 1:0.14
+0.08873024 1:0.15
+0.12595614 1:0.16
+0.15247323 1:0.17
+0.25956145 1:0.18
+0.20040796 1:0.19
+0.19581846 1:0.20
+0.15757267 1:0.21
+0.13717491 1:0.22
+0.19020908 1:0.23
+0.19581846 1:0.24
+0.20091790 1:0.25
+0.16879143 1:0.26
+0.18510964 1:0.27
+0.20040796 1:0.28
+0.29576747 1:0.29
+0.43396226 1:0.30
+0.53391127 1:0.31
+0.52116267 1:0.32
+0.48546660 1:0.33
+0.49209587 1:0.34
+0.54156043 1:0.35
+0.59765426 1:0.36
+0.56144824 1:0.37
+0.58592555 1:0.38
+0.52983172 1:0.39
+0.50178480 1:0.40
+0.52626211 1:0.41
+0.58286588 1:0.42
+0.64660887 1:0.43
+0.68077511 1:0.44
+0.74298827 1:0.45
+0.64864865 1:0.46
+0.67261601 1:0.47
+0.65782764 1:0.48
+0.69811321 1:0.49
+0.63029067 1:0.50
+0.61601224 1:0.51
+0.63233044 1:0.52
+0.65323814 1:0.53
+0.65323814 1:0.54
+0.67363590 1:0.55
+0.67006629 1:0.56
+0.51555329 1:0.57
+0.50892402 1:0.58
+0.33299337 1:0.59
+0.36206017 1:0.60
+0.43090260 1:0.61
+0.45996940 1:0.62
+0.56348802 1:0.63
+0.54920959 1:0.64
+0.48393677 1:0.65
+0.48495665 1:0.66
+0.46965834 1:0.67
+0.45181030 1:0.68
+0.45843957 1:0.69
+0.47118817 1:0.70
+0.51555329 1:0.71
+0.58031617 1:0.72
+0.55481897 1:0.73
+0.56297807 1:0.74
+0.56603774 1:0.75
+0.57929628 1:0.76
+0.64762876 1:0.77
+0.66241713 1:0.78
+0.69301377 1:0.79
+0.65119837 1:0.80
+0.68332483 1:0.81
+0.66598674 1:0.82
+0.73890872 1:0.83
+0.73992861 1:0.84
+0.84242733 1:0.85
+0.91330954 1:0.86
+0.88016318 1:0.87
+0.90719021 1:0.88
+0.93115757 1:0.89
+0.93115757 1:0.90
+0.91942886 1:0.91
+0.92911780 1:0.92
+0.95665477 1:0.93
+0.95002550 1:0.94
+0.96940337 1:0.95
+1.00000000 1:0.96
+0.89801122 1:0.97
+0.90311066 1:0.98
+0.90362060 1:0.99
+0.83477817 1:1.0

data/mllib/sample_isotonic_regression_libsvm_data.txt

@@ -691,6 +691,76 @@ The implementation matches the result from R's survival function
 </div>
 
 
+## Isotonic regression
+[Isotonic regression](http://en.wikipedia.org/wiki/Isotonic_regression)
+belongs to the family of regression algorithms. Formally isotonic regression is a problem where
+given a finite set of real numbers `$Y = {y_1, y_2, ..., y_n}$` representing observed responses
+and `$X = {x_1, x_2, ..., x_n}$` the unknown response values to be fitted
+finding a function that minimises
+
+`\begin{equation}
+  f(x) = \sum_{i=1}^n w_i (y_i - x_i)^2
+\end{equation}`
+
+with respect to complete order subject to
+`$x_1\le x_2\le ...\le x_n$` where `$w_i$` are positive weights.
+The resulting function is called isotonic regression and it is unique.
+It can be viewed as least squares problem under order restriction.
+Essentially isotonic regression is a
+[monotonic function](http://en.wikipedia.org/wiki/Monotonic_function)
+best fitting the original data points.
+
+We implement a
+[pool adjacent violators algorithm](http://doi.org/10.1198/TECH.2010.10111)
+which uses an approach to
+[parallelizing isotonic regression](http://doi.org/10.1007/978-3-642-99789-1_10).
+The training input is a DataFrame which contains three columns
+label, features and weight. Additionally IsotonicRegression algorithm has one
+optional parameter called $isotonic$ defaulting to true.
+This argument specifies if the isotonic regression is
+isotonic (monotonically increasing) or antitonic (monotonically decreasing).
+
+Training returns an IsotonicRegressionModel that can be used to predict
+labels for both known and unknown features. The result of isotonic regression
+is treated as piecewise linear function. The rules for prediction therefore are:
+
+* If the prediction input exactly matches a training feature
+  then associated prediction is returned. In case there are multiple predictions with the same
+  feature then one of them is returned. Which one is undefined
+  (same as java.util.Arrays.binarySearch).
+* If the prediction input is lower or higher than all training features
+  then prediction with lowest or highest feature is returned respectively.
+  In case there are multiple predictions with the same feature
+  then the lowest or highest is returned respectively.
+* If the prediction input falls between two training features then prediction is treated
+  as piecewise linear function and interpolated value is calculated from the
+  predictions of the two closest features. In case there are multiple values
+  with the same feature then the same rules as in previous point are used.
+
+### Examples
+
+<div class="codetabs">
+<div data-lang="scala" markdown="1">
+
+Refer to the [`IsotonicRegression` Scala docs](api/scala/index.html#org.apache.spark.ml.regression.IsotonicRegression) for details on the API.
+
+{% include_example scala/org/apache/spark/examples/ml/IsotonicRegressionExample.scala %}
+</div>
+<div data-lang="java" markdown="1">
+
+Refer to the [`IsotonicRegression` Java docs](api/java/org/apache/spark/ml/regression/IsotonicRegression.html) for details on the API.
+
+{% include_example java/org/apache/spark/examples/ml/JavaIsotonicRegressionExample.java %}
+</div>
+<div data-lang="python" markdown="1">
+
+Refer to the [`IsotonicRegression` Python docs](api/python/pyspark.ml.html#pyspark.ml.regression.IsotonicRegression) for more details on the API.
+
+{% include_example python/ml/isotonic_regression_example.py %}
+</div>
+</div>
+
+
 
 # Decision trees
 

docs/ml-classification-regression.md

@@ -157,7 +157,7 @@ The history server can be configured as follows:
       If enabled, access control checks are made regardless of what the individual application had
       set for <code>spark.ui.acls.enable</code> when the application was run. The application owner
       will always have authorization to view their own application and any users specified via
-      <code>spark.ui.view.acls</code> and groups specified via <code>spark.ui.view.acls.groups<code>
+      <code>spark.ui.view.acls</code> and groups specified via <code>spark.ui.view.acls.groups</code>
       when the application was run will also have authorization to view that application.
       If disabled, no access control checks are made.
     </td>